PROC STDRATE Error

bhong — Wed, 29 May 2019 05:28:48 GMT

Hi, I'm trying to use PROC STDRATE to get standardized mortality rates. I have two datasets, one of them being the reference dataset. I've tried to run the following code, but get various warning and error messages. I have no idea what I'm doing wrong, so any help would be great!

proc stdrate data = mort_rate_race
				refdata = ref_race
				method = direct
				stat = rate (mult = 100)
				effect = ratio
				plots(only) = effect
				;
	population event = total_death total = total_risk_period;
	reference total = total_pop;
	strata race_new / effect;
run;

However, I get the following warning/error messages:

WARNING: You must enable ODS graphics before requesting plots.
WARNING: The GROUP= option in the POPULATION statement is required for the EFFECT option. The option is not used.
WARNING: The stratum in the REFDATA= data set, race_new=	Asian, does not have a matched stratum in the DATA= data set.
WARNING: The stratum in the REFDATA= data set, race_new=	Black, does not have a matched stratum in the DATA= data set.
WARNING: The stratum in the REFDATA= data set, race_new=	Hispanic, does not have a matched stratum in the DATA= data set.
WARNING: The stratum in the REFDATA= data set, race_new=	Other, does not have a matched stratum in the DATA= data set.
WARNING: The stratum in the REFDATA= data set, race_new=	White, does not have a matched stratum in the DATA= data set.
ERROR: The reference total is missing in the stratum race_new=Asian.

I'm not sure why, as I've looked at my two datasets and everything seems to be there. Here are the two datasets:

Re: PROC STDRATE Error

FreelanceReinh — Wed, 29 May 2019 10:34:33 GMT

Hi @bhong,

I'm not very familiar with PROC STDRATE, but a quick comparison with Example 110.1 Comparing Directly Standardized Rates from the procedure documentation reveals an important difference: There, they compare two populations (Alaska and Florida) which is why they need to standardize (using data for the U.S. as the reference population). Your MORT_RATE_RACE dataset, however, contains data for only one population. This means that all options involving "EFFECT" are not applicable because they apply "only when METHOD=DIRECT with two study populations ... are required" (documentation). So, you should delete or comment out the EFFECT= and PLOTS= options of the PROC STDRATE statement as well as the EFFECT option of the STRATA statement.

Also, you should make sure that the values in variable TOTAL_POP have the same unit (e.g. person-years) as those in variable TOTAL_RISK_PERIOD.

Moreover, you have a data issue: It looks like variable RACE_NEW in dataset REF_RACE contains tabs, leading blanks or other white space characters, which cause the mismatches mentioned in five of the warnings (e.g. ' Asian' does not match 'Asian'). So, you need to resolve these issues, e.g., with a DATA step like

data ref_race;
set ref_race;
race_new=compress(race_new,,'ka');
run;

or, even better, in the step that creates REF_RACE in the first place. If REF_RACE values in MORT_RATE_RACE also have leading blanks, you need to remove them in the same way.

Finally, to enable ODS graphics (for the implied default plot request PLOTS=RATE) use the ODS GRAPHICS statement.

ods graphics;

proc stdrate data = mort_rate_race
    refdata = ref_race
    method = direct
    stat = rate (mult = 100)
    ;
  population event = total_death total = total_risk_period;
  reference total = total_pop; /* <-- please check the unit of TOTAL_POP */
  strata race_new;
run;

The above code should run without errors or warnings. It's your responsibility to check if the results make sense.

topic PROC STDRATE Error in Statistical Procedures

PROC STDRATE Error

Re: PROC STDRATE Error